Deliverable D5.5 Model evaluation - KIT · Project No: 603502 DACCIWA...

Project No: 603502

DACCIWA

"Dynamics-aerosol-chemistry-cloud interactions in West Africa"

Deliverable

D5.5 Model evaluation

Due date of deliverable: 30/11/2018

Completion date of deliverable: 06/12/2018

Start date of DACCIWA project: 1st December 2013 Project duration: 60 months

Version: [V1.0]

File name: [D5.5_Model_evaluation_DACCIWA_v1.0.pdf]

Work Package Number: 5

Task Number: 5

Responsible partner for deliverable: ECMWF

Contributing partners: UREAD

Project coordinator name: Prof. Dr. Peter Knippertz

Project coordinator organisation name: Karlsruher Institut für Technologie

The DACCIWA Project is funded by the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no 603502.

DACCIWA 603502 2/22

D5.5_Model_evaluation_DACCIWA_v1.0 www.dacciwa.eu

Dissemination level

PU Public x

PP Restricted to other programme participants (including the Commission Services)

RE Restricted to a group specified by the consortium (including the Commission Services)

CO Confidential, only for members of the consortium (including the Commission Services)

Nature of Deliverable

R Repo1 x

P Prototype

D Demonstrator

O Other

Copyright

This Document has been created within the FP7 project DACCIWA. The utilization and release of this

document is subject to the conditions of the contract within the 7th EU Framework Programme. Project

reference is FP7-ENV-2013-603502.

DACCIWA 603502 3/22


DOCUMENT INFO

Authors

Author Beneficiary Short

Name E-Mail

Peter Hill UREAD [email protected]

Angela Benedetti ECMWF [email protected]

Anke Kniffka KIT [email protected]

Christine Chiu UREAD [email protected]

Peter Knippertz KIT [email protected]

Changes with respect to the DoW

Issue Comments

Dissemination and uptake

Target group addressed Project internal / external

Scientific community Internal and external

Document Control

Document version # Date Changes Made/Comments

V0.1 24.10.2018 Template with basic structure

V0.2 06.11.2018 First draft

V0.3 15.11.2018 Version with input of all authors for approval by the General Assembly

V1.0 06.12.2018 Final version approved by the General Assembly

mailto:[email protected]

DACCIWA 603502 4/22


Table of Contents 1 Introduction .............................................................................................................................. 5

2 Observations of cloud and aerosol radiative effects ................................................................. 5

2.1 Cloud radiative effects ...................................................................................................... 5

2.2 Aerosol radiative effects .................................................................................................... 9

3 Evaluation of model radiation ................................................................................................. 12

3.1 DACCIWA campaign forecasts ....................................................................................... 12

3.2 Global atmospheric models............................................................................................. 13

3.3 Further analysis of climate models .................................................................................. 15

4 Effect of radiative processes on circulation ............................................................................ 17

4.1 Energy budget analysis ................................................................................................... 17

4.2 Effect of SW irradiance perturbations on regional precipitation and circulation ............... 18

5 Summary and conclusions ..................................................................................................... 19

6 References ............................................................................................................................ 20

DACCIWA 603502 5/22


1 Introduction

Understanding of cloud radiative effects (CREs) in the southern West Africa region is currently

limited. The complex cloud climatology with frequent multilayer clouds in this region (Stein et al.,

2011) makes it difficult to identify cloud types and to attribute radiative effects to different cloud

regimes. A lack of surface-based cloud observations (Knippertz et al., 2015a) and uncertain

aerosol–cloud interactions (Knippertz et al., 2015b) further limit understanding of clouds in this

region.

Compared to CRE, direct aerosol direct radiative effects (DARE) are perhaps even less well

understood over this region. While this lack of understanding is partly due to a lack of surface-

based aerosol measurements, the prevalence of cloud in this region poses additional challenges in

measuring aerosol properties and quantifying ADRE. These challenges include infrequent satellite

aerosol retrievals, since the retrieval methods require clear-sky conditions to avoid signal

contaminations from clouds. Since ADRE strongly depends on surrounding cloud conditions,

which are uncertain themselves, it is even more challenging to quantify ADRE than CRE.

Within the DACCIWA project, we have estimated cloud and aerosol radiative effects using multiple-

satellite dataset built from Task 5.1 and 5.3, evaluated against field campaign data in Task 5.2 and

5.4. Combining these with simulations from operational Numerical Weather Prediction (NWP)

models and climate models (in collaboration with WP7), we can develop a new understanding of

the interactions and potential feedbacks between cloud and aerosol radiative effects and

atmospheric circulation across various temporal and spatial scales.

The main objective of the report is 1) to quantify the occurrence of various cloud types and their

associated radiative effects in the DACCIWA region during the monsoon season; 2) to report

aerosol direct radiative effects; and 3) to discuss model errors in radiation and their impacts on

energy budget, cloud type transition and circulation. Note that aerosol indirect effects are

discussed in D4.4 and thus will not be included here. Also, for consistency, this report focuses on

the region centred upon 5–10°N and 8°W–8°E. We shall refer to this region hereafter as the

“DACCIWA region”.

2 Observations of cloud and aerosol radiative effects

This section documents work completed within the DACCIWA project, towards an improved

understanding of how clouds and aerosols affect radiation budget over the DACCIWA region.

Such an understanding is required to ultimately reduce persistent radiation errors found in

atmospheric models in this region.

2.1 Cloud radiative effects

Radiative effects of various cloud types in the DACCIWA region are quantified using the state-of-

the-art satellite datasets and radiative transfer. The CERES–CloudSat–CALIPSO–MODIS

(CCCM) dataset (Kato et al., 2010; Kato et al., 2011; Ham et al., 2017), which combines

observations from spaceborne active and passive instruments, provides detailed cloud profiles

required for radiative transfer calculations. These cloud profiles were based on measurements of

polar-orbiting satellites that cross the equator at approximately 0130 and 1330 local time (LT). The

release B1 dataset is available from July 2006 to April 2011 inclusive. Since we focus on monsoon

seasons in June–September to coincide with research activities of the DACCIWA project

(Knippertz et al., 2015a; Hill et al., 2016; Hannak et al., 2017), the choice of the time period leads

to a total data length of 19 months.

DACCIWA 603502 6/22


Radiative fluxes at the top of the atmosphere (TOA), at the surface, and heating rates in the

atmosphere are calculated using the CCCM dataset as input to the Suite of Community Radiative

Transfer Codes (SOCRATES) two-stream radiation scheme (Edwards & Slingo, 1996). The fluxes

are evaluated against coincident the Clouds and the Earth's Radiant Energy System (CERES;

Wielicki et al., 1996) measurements, showing good agreement with Pearson correlation coefficients

of 0.95 for the TOA shortwave (SW) irradiances, 0.85 for the 1330 LT TOA longwave (LW)

irradiances and 0.92 for the 0130 LT TOA LW irradiances. Diurnal mean irradiances in the LW are

estimated by averaging the two CCCM sample times, while diurnal mean irradiances in the SW are

estimated by running the calculations using 24 different solar zenith angles. More details can be

found in Hill et al. (2018).

Following the scheme described in Tselioudis et al. (2013), each CCCM profile is classified as a

cloud type based on its vertical structure. Pressure thresholds of 680 and 440 hPa are used to

determine whether the profile contains one or more of low- (L), mid- (M), or high-level (H) cloud

and whether cloud in different layers is connected or not. As illustrated in Fig. 1, this classification

results in 13 different scene types, including clear sky and 12 cloud types.

Fig. 1. Schematic illustrating how each cloudy CCCM profile is assigned to one of 12 cloud types

based on the cloud vertical structure. L, M and H represent low-, mid- and high- clouds,

respectively. Cloud occurring in multiple layers is denoted by a letter for each layer it occurs in,

while the letter x is used to denote when cloud extends across the pressure boundaries. From Hill

et al. (2018).

To provide further insight into how different cloud types affect the regional energy budget, the

contribution to the total cloud radiative effect (CRE) from each cloud type is further decomposed

into its frequency of occurrence and mean coincident cloud radiative effect (CCRE; i.e., the mean

radiative effect calculated using only the CCCM group profiles that correspond to that cloud type).

The frequency of occurrence of the different cloud types in Fig. 2 shows that the DACCIWA region

has infrequent clear sky and is very cloudy, in agreement with existing cloud climatology (Hill et al.,

2016). The most common cloud types are 1L, 1H, and HL, but 8 of the 12 cloud types occur at

least 5% of the time in this region, indicating a much more diverse set of cloud types than those

found in many other parts of the globe (Tselioudis et al., 2013; Bodas-Salcedo et al., 2016).

Multilayer clouds, where distinct clouds occur simultaneously in multiple layers, occur rather

frequently (42% during the day and 46% during the night), representing a further source of

complexity for understanding CRE.

DACCIWA 603502 7/22


Fig. 2. Mean frequency of occurrence of each cloud type in the CCCM product over the

DACCIWA region in June–September, 2006–2010. Cloud frequency of occurrence at 1330 and

0130 LT are normalized separately. Uncertainty resulting from sampling is illustrated by the error

bars, which show the 95% confidence interval based on bootstrapping. From Hill et al. (2018).

Over the DACCIWA region, the CRE of different clouds depends on the particular irradiance in

question (Hill et al., 2018). The CRE is complex, since almost all the cloud types make a non-

negligible contribution to some aspect of the radiation budget. As shown in Fig. 3a, the mean SW

CCRE of each cloud type at TOA is strongly linked to the number of layers it extends through, i.e.,

the physical thickness of cloud layers. This is not surprising, because physical thickness is

correlated with water path and optical depth (Wang et al., 2000). The HxMxL cloud type, which

extends into three layers and is likely to be deep convection, has the largest mean SW CCRE (476

W m−2 at 1330 LT). Those cloud types that extend between two layers have the next largest mean

SW CCRE with values ranging from 275 to 297 W m−2 at 1330 LT. Clouds that occur separately in

one or more layers have CCRE values ranging from 150 to 187 W m−2 at 1330 LT. The mean LW

CCRE at TOA (Fig. 3b) is of a smaller magnitude compared to SW CCRE (Fig. 3a) for almost all

cloud types, with isolated high cloud being the exception. Unlike SW, the magnitude of LW CCRE

at TOA is mainly determined by cloud-top temperature, and thus closely linked to the presence of

high cloud (e.g., 1H and HxMxL cloud types).

DACCIWA 603502 8/22


Fig. 3. The mean (a) SW and (b) LW mean CCRE at the TOA over the DACCIWA region over

June–September, 2006–2010. Bars labelled 0130 and 1330 LT correspond to calculations based

on the night-time and daytime satellite overpasses, respectively. The diurnal approximation shown

in (a) is based on averaging calculations that use the daytime CCCM data and a range of solar

zenith angles. Uncertainty resulting from errors in our calculations is illustrated by the constrained

calculations, which exclude CCCM group profiles where the SOCRATES–CERES TOA differences

are large. Error bars show the 95% confidence interval based on bootstrapping. From Hill et al.

(2018)

Figure 4 shows the approximate diurnal mean total (i.e., SW + LW) cloud radiative effects. The

diurnal mean total irradiances tend to be small because of cancellation between LW and SW

CREs. For some cloud types, uncertainty is large (up to ±7 W m−2) at the TOA and surface, but the

uncertainty is generally much smaller for fluxes into the atmosphere. At the TOA, the 1L cloud

type has the largest magnitude net CRE, as the decrease in net downwelling SW irradiance

resulting from low clouds is much larger than the increase in net downwelling LW irradiance. Most

other cloud types also have a negative effect on the TOA net downwelling irradiance, although for

many cloud types this is not certain. Isolated high cloud (1H) is the only cloud type that definitely

leads to an increase in the net TOA irradiance. At the surface, all cloud types reduce the net

downwelling irradiance, because the reduction in SW radiation reaching the surface is larger than

the increase in downwelling LW radiation. 1L leads to a small reduction in the flux into the

atmosphere, but all other cloud types increase the flux into the atmosphere.

For low clouds, a particular focus of the DACCIWA project, we have found that the 1L cloud type

occurs with a frequency of 17 % at 1330 LT and 7 % at 0130 LT (Fig. 2). Low clouds that are

beneath higher clouds occur more frequently, with a frequency of 31 % at 1330 LT and 29 % at

DACCIWA 603502 9/22


0130 LT. Note that these low clouds beneath higher clouds cannot easily be detected by passive

satellite observations. Moreover, although these low clouds beneath higher clouds may not always

have a particularly large effect on the TOA radiation budget, they do have a significant impact on

the surface radiation budget (e.g. Fig. 9 in Hill et al. [2018]).

To summarize, the DACCIWA region experiences many different cloud types; no single cloud type

dominates in terms of either frequency of occurrence or radiative effect. The most frequent cloud

types are 1L, 1H, HL, and HxMxL, which have frequencies of 12%, 14%, 19%, and 10%,

respectively. Contributions from different cloud types to the regional mean cloud radiative effect

depend not only on their frequencies, but also on their mean CCRE, which is linked to cloud

thickness in the SW and cloud-top and cloud-base height in the LW.

Fig. 4. Contribution to the diurnal mean total (i.e., SW + LW) CRE from each cloud type for June–

September 2006–2010 over the DACCIWA region, based on radiative transfer calculations. Error

bars show the combined uncertainty resulting from the diurnal mean approximation, the

constrained calculation (which exclude CCCM group profiles where the SOCRATES–CERES TOA

differences are large), and the limited sampling. These uncertainties are calculated separately for

the SW and LW and are combined in quadrature. From Hill et al (2018).

2.2 Aerosol radiative effects

Similarly to cloud radiative effects, aerosol radiative effects are estimated by combining CCCM

data with SOCRATES radiative transfer calculations. Seven common aerosol species are

provided by CCCM, including soluble and insoluble particles, small and large dust particles, sulfuric

acid, sea salt, and soot, as defined in the Optical Properties of Aerosols and Clouds (OPAC)

climatology (Hess et al., 1998). The extinction CCCM attributes to each species depends on

CALIPSO measurements and the Model of Atmospheric Transport and Chemistry (MATCH; Rasch

et al., 1997). The extinction, single scattering albedo, and asymmetry of these aerosol species are

spectrally dependent and parameterized in SOCRATES as a function of aerosol mass mixing ratio,

as described in Cusack et al. (1998). Since the CCCM dataset does not provide mass mixing ratio

DACCIWA 603502 10/22


directly, we used the inverse of the SOCRATES parameterization to derive profiles of aerosol

mass mixing ratios from the CCCM aerosol extinction profiles for each species. These aerosol

mass mixing profiles are then input to SOCRATES for radiative transfer calculations. Additionally,

the diurnal means of radiative effects at TOA are estimated by averaging LW irradiances at 1330

and 0130 LT for the LW, and by multiplying the SW irradiances at 1330 LT by a factor of 0.396 for

the SW. This multiplying factor was estimated by comparing the SW irradiances at 1330 LT to the

mean of the hourly SW irradiances at TOA.

Fig. 5. Zonal mean aerosol optical depth at 500 nm for the DACCIWA region, from the CCCM

dataset for June–September, 2006–2010. The solid black line indicates the total aerosol optical

depth from all aerosol species, while each of the broken lines indicates the optical depth due to

one of the seven aerosol species included in CCCM.

Figure 5 shows the zonal mean aerosol optical depth attributed to each species in CCCM. The

total aerosol optical depth is ~0.32 near the coast and decreases to ~0.3 inland. This value of 0.3

generally agrees with the mean AOD measured by the NASA Aerosol Robotic Network

(AERONET) in June-July 2016, as described by Mollard et al. (2018). Although the total optical

depth is rather constant, individual aerosol species show different variability with latitude. The

optical depth of small dust aerosols remains fairly constant with latitude, but the optical depth of

large dust particles increases with latitude, as distance to the major dust source decreases. The

sulphate aerosol corresponds to stratospheric background aerosol and thus it is unsurprising that

its optical depth shows little variation with latitude. Extinction due to sea salt is negligible at most

latitudes, with larger (but still small) optical depth at the coast. The optical depth of soot is also

very small, with largest values again at the coast. Soluble aerosol corresponds to pollution and as

a result the optical depth is largest near the coast where the large cities (i.e., the main emission

sources) are located and decreases with latitude. Finally, the optical depth of insoluble aerosols,

which correspond to soil particles, is negligible at all latitudes. Since DACCIWA aircraft

measurements show that the regional pollution is dominated by biomass burning (Haslett et al., in

preparation), the CCCM product might have underestimated the optical depth of soot.

The zonal mean direct aerosol radiative effect (DARE) for both the SW and LW at both TOA and

surface is shown in Fig. 6. Similar to CRE, the DARE is defined as the difference of irradiances

computed with and without aerosols. Since aerosol radiative effects are influenced by the

DACCIWA 603502 11/22


associated cloud conditions, we show separate estimates of the aerosol radiative effects using

both all-sky (i.e. including the observed cloud) and clear-sky (i.e. removing all cloud) calculations.

Fig. 6. Zonal mean aerosol direct radiative effects for the DACCIWA region in June–September

2006–2010, based on CCCM data and SOCRATES calculations with and without aerosols. “ All-

sky” values (solid lines) are computed including clouds in radiative transfer, while “clear-sky”

values (broken lines) are computed excluding clouds in radiative transfer calculations with and

without aerosols.

Beginning with the LW, we see that aerosols lead to an increase in the net downwelling LW

irradiance at both the surface and TOA. All-sky DARE increases from ~1 W m–2 near the coast to

~2 and ~4 W m–2 near 10°N at the TOA and surface, respectively. If clouds are excluded from the

calculations (i.e., clear-sky estimates), LW DARE are consistently ~3 W m–2 larger at both the TOA

and surface for all latitudes, remaining the increase.

The increase with latitude in LW DARE is thought to be due to the changes in aerosol species.

Recall that Fig. 5 presents optical depth at 500 nm, a wavelength in the SW. Let us focus on two

aerosol species in Fig. 5: one is large dust particles that show the largest increase in optical depth

with latitude, and the second is soluble particles that show the largest decrease in optical depth

with latitude. The opposite changes with altitude in these two species help maintain the total

optical depth at 500 nm to be nearly constant. However, the extinction per unit mass of soluble

particles in the LW is at least 10 times smaller than that at 500 nm, while the extinction per unit

mass of large dust particles in the LW is similar to that at 500 nm. As a result, in the LW, the

increase in extinction with latitude due to increased concentrations of large dust particles is larger

than the decrease in extinction due to decreased concentrations of soluble particles. This leads to

an increase in the total extinction in the LW increases with latitude, and the increase LW DARE

with latitude seen in Fig. 6.

In the SW, aerosols lead to a decrease in the net downwelling SW irradiance at both the surface

and TOA. At the surface, the clear-sky DARE is approximately –15 W m–2 near the coast and

decreases to about –13.5 W m–2 at higher latitudes. This is a direct consequence of the changes

in the total aerosol optical depth with latitude. At the TOA, the DARE is smaller than at the surface,

as absorption by aerosols reduces the amount of downwelling SW irradiance, but does not

necessarily change the upwelling SW irradiance. The magnitude of the clear-sky SW DARE at

DACCIWA 603502 12/22


TOA decreases with latitude between the coast and ~7 °N. This is due to an increase in

absorption by aerosols as the concentrations of more scattering aerosols (e.g. soluble) decrease

with latitude, while the concentrations of more absorbing aerosols (e.g. large dust) increase with

latitude, as shown in Fig. 5. The all-sky surface SW DARE is smaller in magnitude, as clouds

extinguish some of the radiation that is extinguished by aerosols in the clear-sky case.

Although both the mean AOD and mean SW incoming irradiance used in our calculations are

larger than the global averages, the zonal mean calculated clear-sky values for the SW DARE at

TOA are similar to estimates of the global mean over land at both the TOA (between –1.7 and –5.5

W m–2) and surface (between –5.1 and –13.5 W m–2) as estimated by Yu et al. (2006). Our

calculated zonal mean is also very close to the value of 5.82 W m–2 estimated by Mollard et al.

(2018) using AERONET aerosol properties and assuming clear-sky at Save (note the difference in

sign is due to comparing the DARE on net downwelling TOA SW irradiances to the DARE for

upwelling TOA SW irradiances). However, the SW surface DARE is notably ~10 W m–2 smaller in

magnitude than the value of –26.08 W m–2 calculated by Mollard et al. (2018). About 5 W m–2 of

this difference is because the results presented here are net downwelling (i.e. down-up) as

opposed to downwelling. Since the TOA DARE values are in good agreement, the other 5 W m–2

difference must be due to additional atmospheric absorption from aerosols in the AERONET

calculations.

3 Evaluation of model radiation

3.1 DACCIWA campaign forecasts

As part of the forecast evaluation reported in D7.3, several model forecasts were verified using

Outgoing Longwave Radiation (OLR) using measurements from Geostationary Earth Radiation

Budget (GERB) instrument processed at the University of Reading.

GERB is the geostationary earth radiation budget instrument onboard the EUMETSAT’s Meteosat

Second Generation (MSG) satellites (Harries et al., 2005). It is a broadband radiometer designed

to measure the total emitted and solar reflected radiances over the earth with high temporal

resolution (5 min) and 50 km spatial resolution. From the measurements, the top of atmosphere

radiation, divided in short and longwave contributions can be estimated to give an accurate

estimation of the earth’s radiation budget (or rather the budget in MSG’s field of view which covers

almost half of the earth’s surface).

GERB measurements were averaged of the preceding three hours, calculated from measurements

every 15 minutes. To avoid biasing the fluxes due to diurnal sampling, if any GERB data in the

three-hour window was missing, the three-hour average was set to missing. GERB solar flux

suffers from sun-glint over water surfaces at low zenith angles, which means that these are often

missing. Note that when sun glint occurs, the daily means are biased low because it occurs at low

zenith angles. GERB solar flux is only measured for zenith angles less than 80°. The data were

extended to 104.5° using CERES twilight measurements and interpolation as described in Hill et

al. (2016).

Figure 7 shows OLR time series and errors for the pre-onset and post-onset periods, as defined by

Knippertz et al. (2017). Most forecast models have a lower bias in the post-onset period than in the

pre-onset but large centred root mean square error (CRMSE) in both periods. A small exception is

the COSMO model that displays a large bias in the post-onset period, possibly connected with an

underestimation of cirrus outflow resulting from deep convection.

DACCIWA 603502 13/22


Fig. 7. Box-average OLR from the various models compared with GERB observations during the

DACCIWA campaign. Left: Time-series. Right: Error diagrams with contributions from systematic

and unsystematic errors to the Root Mean Square Error with the different times of day combined.

The latter is referred to as “centred root mean square error (CRMSE)”. Top: pre-onset period;

Bottom: post-onset period. (From Fig. 13, D7.3.)

3.2 Global atmospheric models

A detailed evaluation of the DACCIWA climate models was presented in D7.4. We present results

from an evaluation of the representation of the West African monsoon (WAM) system in widely

used global and regional climate and seasonal model datasets following a similar strategy as in

D7.3 for weather forecasting. Here we present a sub-set of D7.4 focusing on TOA radiative fields.

Considered model datasets include (a) long-term climate simulations from the Coupled Model

Intercomparison Project Phase 5 (CMIP5) coordinated by the Intergovernmental Panel on Climate

Change (IPCC), (b) climate simulations conducted as part of the Years of Tropical Convection

(YoTC; Waliser et al., 2012), which provides additional tendency terms and diurnally resolved data,

(c) own climate simulations using the ECHAM6 (run by ETHZ) and Unified Model (run by MO), and

(d) seasonal forecasts using the ECMWF Integrated Forecast System (IFS). Outgoing longwave

and shortwave radiation from forecast models were evaluated against GERB products. We used

the “Standard High-resolution Image” (Dewitte et al., 2008), in which the data are processed using

cloud observations from SEVIRI to convert radiance measurements to radiative fluxes at

approximately 10 km resolution, every 15 min.

As shown in Fig. 8, the highest values of OLR from GERB are found in the southern west corner of

the DACCIWA box, the smallest ones in the southern right corner, while a triangle-shaped region

with medium values can be found in between. This pattern is very well repeated by the IFS runs,

DACCIWA 603502 14/22


and most other models show at least the maximum in approximately the right position. The

absolute values are too high from all models but IFS. Hence, the radiative effect of clouds is not

modelled in the right way. The GERB outgoing shortwave radiation (OSR) field places the smallest

values at the southern edge in the centre with larger values shortly northwards from it. All models

show the same structure but the absolute values vary in a range of 190 W m–2 which is much more

than for OLR (55 W m–2).

Fig. 8. Outgoing longwave radiation (upper panel) and outgoing shortwave radiation (lower panel)

at top of the atmosphere as seasonal mean for the YOTC models, the ECHAM and the IFS runs

and the UM compared to GERB measurements from 2006–2015. Please note that for IFS only net

radiation was available, therefore shortwave fields cannot be shown and for the longwave fields the

incoming longwave radiation was assumed to be exactly zero.

The influence of clouds on outgoing shortwave and longwave radiation varies from model to model.

The UM had by far the lowest amount of liquid water and not too much ice water, therefore it is not

surprising that the seasonal mean of OLR for the UM is largest compared to the other models. For

the other models the effect of clouds is not easy to describe without additional information, e.g.

NAVGEM does not have a lot of liquid water or ice but at the same time not a very high OLR. This

could either mean that the diurnal cycle is very strong and no general conclusions can be drawn

from comparing a figure for 12 UTC with all-day averages, or that the microphysical properties of

the clouds contribute strongly to the observed variation.

DACCIWA 603502 15/22


3.3 Further analysis of climate models

Having examined radiation errors in models, this section examines TOA irradiances in greater

detail in climate models. In particular, we analyse interannual variability and diurnal cycles in the

climate models and identify the causes of errors in these models by considering all-sky and clear-

sky irradiances separately. Further details concerning this analysis are available from Hill et al

(2016).

We base this analysis on simulations from CMIP5 (Taylor et al., 2012). As this study is focused on

atmospheric processes, we use Atmospheric Model Intercomparison Project (AMIP) simulations

that employ predetermined realistic sea surface temperatures and sea ice. Additionally, since the

DACCIWA region exhibits notable diurnal cycles of cloud and hence radiation (Stein et al., 2011;

van der Linden et al., 2015), we also focus on those AMIP5 models for which 3‐hourly output is

available: CanAM4, CNRM‐CM5, HadGEM2‐A, and MRI‐CGCM3. We focus on the months of

June and July, using 1) data from 2008 for which 3 hourly model output is available for diurnal

cycle comparisons, and 2) all available years from 2002 onwards for diurnal mean comparisons.

We use the CERES data set to evaluate these climate models. CERES instruments are limited to

two polar‐orbiting satellites and as a result only sample four points in the diurnal cycle. The

CERES instrument on the AQUA satellite takes radiation measurements over the DACCIWA

region at approximately 0130 and 1330 LT, while the CERES instrument on board the TERRA

satellite takes measurements over the DACCIWA region at approximately 1030 and 2230 LT. To

obtain better sampling of the diurnal cycle, we use the synoptic radiative fluxes and clouds

(CERES‐SYN) product (Doelling et al., 2013), which capitalizes on geostationary imagers to

estimate the diurnal variability between CERES observations. For consistency with the climate

models, we use CERES-SYN data from June–July 2008 for diurnal cycle comparisons and June-

July 2002–2014 for diurnal mean comparisons.

Figure 9 shows the June-July mean all-sky and clear-sky irradiances for the CERES-SYN product

and the four AMIP models. For the TOA SW irradiances, interannual variability in the each of the

models is large. The interannual variability in the CERES-SYN product is smaller. Two of the

AMIP models consistently overestimate the TOA outgoing SW irradiance compared to CERES-

SYN, while two underestimate it. For TOA LW irradiances, the modelled interannual variability is

again larger than the observed interannual variability. Generally the climate models overestimate

the TOA outgoing LW irradiance, but only one of the climate models (HadGEM2-A) overestimates

the TOA outgoing LW irradiance by a significant amount.

Examining clear-sky irradiances can help determine whether the all-sky irradiance errors are due

to cloud or other processes such as aerosols, water vapour, or the surface albedo. However, it is

important to bear in mind that the models and satellite observations determine clear-sky

irradiances quite differently. In the models, clear-sky irradiances are calculated by repeating

radiative transfer calculations without cloud. In the observations, clear-sky irradiances are

estimated by averaging the measurements for pixels that are identified as cloud free. This can

lead to differences in mean clear-sky irradiances as large as 15 W m–2 in the LW (Allan and

Ringer, 2003). Nevertheless, this difference is apparently much smaller than the model bias

shown in Fig. 9. For SW, there is a clear climate model bias in clear-sky conditions: all four AMIP

models have larger irradiances than CERES-SYN. In particular, the clear-sky bias in the MRI-

CGCM3 model can explain most of the all-sky bias. Hill et al. (2016) found that this is primarily

because MRI-CGCM3 has a much larger surface albedo than the other models, highlighting the

need for improved observational surface albedo datasets in this region. For the HadGEM2-A and

CNRM-CM5 models, the clear-sky bias appears to partially offset cloud errors and reduce all-sky

DACCIWA 603502 16/22


TOA SW irradiance errors. LW clear-sky errors diurnal mean errors are much smaller and we

cannot identify any definite biases given the uncertainty related to different definitions of clear-sky.

Fig. 9. June-July mean top of atmosphere radiative fluxes in the DACCIWA region for CERES-SYN

and AMIP5 models. Shading shows spatial variability within the domain. Adapted from Hill et al.

(2016).

Figure 10 compares diurnal cycle of TOA irradiances from the models and CERES-SYN.

Differences in the phase of the diurnal cycles of irradiances cannot be disentangled from the

temporal sampling that the climate models use to reduce the cost of the computationally expensive

radiative transfer schemes. The different climate models use different methods to account for this

reduced temporal sampling, and as a result irradiances output at the same point in time may

represent quite different time periods. Such differences make comparison of the phase of the

diurnal cycles of radiation problematic, so we shall restrict our discussion to the amplitude of the

diurnal cycles.

For SW irradiances, the amplitude of the diurnal cycle is dominated by changes in the solar zenith

angle. Differences in the amplitudes of the SW diurnal cycles are consistent with differences in

diurnal mean irradiances. For LW clear-sky irradiances at TOA, all the models have a larger

amplitude diurnal cycle than CERES-SYN. This may be due to the different methods for

calculating clear-sky irradiances. However, two of the climate models (HadGEM2-A and CanAM4)

have significantly larger amplitude all-sky diurnal cycles than CERES-SYN, which implies there are

issues with the diurnal cycles of cloud in these models.

OLR is generally overestimated in the climate models shown in Figs. 8–10. Further analysis of

some of the climate models shows that the clear-sky OLR predicted by these models is much

closer to the observed values than the all-sky OLR, which points to problems simulating the

radiative effect of high clouds in this region. Many (but not all) of the models also overestimate the

OSR compared to satellite observations, but analysis of clear-sky OSR in some of the models

DACCIWA 603502 17/22


reveals large differences amongst models and between the models and the satellite observations,

which may be able to explain large parts of the all-sky errors.

Fig. 10. Comparison of diurnal cycles for June–July 2008. For shortwave radiation (a and c), the

models are shown at a time that is most consistent with the solar zenith angle used, rather than the

output time. From Hill et al. (2016).

4 Effect of radiative processes on circulation

Radiation errors are of interest because errors in radiative heating cause further errors in

circulation and precipitation. This section aims to identify potential consequences of the model

radiation errors identified in the previous section.

4.1 Energy budget analysis

One of the simplest methods for understanding how radiation is linked to circulation and

precipitation is to construct energy budgets. Under the hydrostatic approximation and assuming

that kinetic energy transport and changes in energy and water storage are negligible as in Muller

and O’Gorman (2011), the time-mean regional atmospheric energy budget may be expressed as

𝐿𝑐𝑃 = 𝑄 + 𝐻𝑑

where 𝐿𝑐 is the latent heat of condensation of water vapor, P is the surface precipitation rate, 𝐻𝑑 is

the atmospheric horizontal divergence of dry static energy and Q is the atmospheric net heating

from radiation and sensible heat.

Satellite datasets provide estimates of precipitation rates, SW column radiative heating and LW

column radiative cooling, while we take estimates of sensible heating, surface evaporation,

DACCIWA 603502 18/22


divergence of moisture and divergence of dry static energy from the ERA-Interim reanalysis (Dee

et al., 2011). Figure 11 illustrates the atmospheric energy budget of the DACCIWA region,

including June-July multi-annual mean estimates for each term. The DACCIWA region is fueled by

a combination of ~110 W m–2 of SW radiative heating, ~170 W m–2 of latent energy, and ~25 W m–2

of sensible heating from the surface. This is balanced by LW net radiative cooling between –188

and –211 W m–2 (depending on dataset) and horizontal divergence of dry static energy of

approximately –105 W m–2.

Fig. 11. Key terms in the atmospheric energy budget for the DACCIWA region. Numbers shown

are June–July means (standard deviations of interannual means in parentheses) for 2000–2015

from the data source indicated by that colour. From Hill et al. (2016).

4.2 Effect of SW irradiance perturbations on regional precipitation and

circulation

Deetz et al. (2018) investigate the total (direct + indirect) aerosol effect using COSMO-ART

simulations. By scaling aerosol concentrations within their simulations, they found that increasing

aerosol concentrations led to decreased surface downwelling SW irradiance, due primarily to the

aerosol direct effect. This in turn affects two features of the regional meteorology: the stratus to

cumulus transition (SCT) that takes place in the DACCIWA region in the course of the day, and the

Atlantic (or maritime) Inflow – a coastal front that develops during daytime and propagates inland

during the evening.

The reduction in downwelling surface SW irradiance affects the SCT through decreased surface

temperatures, which in turn lead to decreased sensible heat fluxes. This in turn reduces the

boundary layer height and consequently the cloud base height, leading to an later break-up of the

stratus cloud. The effect of the reduction in surface SW irradiance on the Atlantic Inflow is also

through decreased surface temperature. This reduction in surface temperature leads to increased

surface pressure, which leads to a decrease in the Atlantic Inflow front velocity.

Additionally, Kniffka et al. (2018) examine how the radiative effect of low-level clouds affects the

regional meteorology, using a regional numerical weather prediction model. They ran simulations

with the Icosahedral Nonhydrostatic (ICON) model over the DACCIWA region and analysed the

DACCIWA 603502 19/22


effect of changing the condensed water content below 700 hPa passed to the radiation scheme

used by their model. In contrast to Deetz et al. (2018), they found significant impacts on the

regional precipitation, with decreasing surface precipitation as the low cloud water content

increased. They argued that this was because increasing low cloud water content leads to

decreased surface SW irradiance, which leads to decreased surface temperature and a more

stable boundary layer, leading to less triggering of convection. They also found large effects on

the diurnal cycle of clouds in the region. However, the changes in low cloud water content had

little impact outside the DACCIWA region in which they were applied.

Both these studies illustrate the sensitivity of the regional meteorology to the radiative processes.

Moreover, although the radiative perturbations in both studies are similar (i.e. primarily changing

the downwelling SW irradiance at the surface), the model responses are quite different. These two

different studies perturbed the radiation budget using different methods, resulting in different

perturbations, and were run for different time periods. Thus, it is unclear whether the differing

model responses are due to differences in the radiative perturbations, differences in the regional

meteorology, differences in the way the models respond to these perturbations, or a combination of

all three. These are just two studies using only two models. It remains to be seen whether these

results can be confirmed in observations and/or generalised to other models. It would also be very

useful to analyse these simulations within the context of the regional energy budget.

5 Summary and conclusions

This report addresses specific objectives that Work Package 5 set out to achieve, namely

quantifying the impacts of low and mid-level clouds (layered and deeper congestus) and aerosols

on the energy budget, evaluating and improving state-of-the-art satellite cloud/aerosol retrievals

and models, and analyzing the effect of cloud radiative forcing on the West Africa monsoon

circulation.

We have found that low clouds that are beneath higher clouds occur frequently, approximately

30% both at daytime and nighttime. In contrast, single-layer low clouds occur much less frequently

and appear to have a more evident daytime (17%) and nighttime (7%) difference. Similar to other

types of clouds, single-layer low cloud has a moderate negative radiative effect at TOA in the SW

and a small positive effect in the LW. Due to the compensation between SW and LW, the diurnal

mean of the total radiative effect of single-layer low clouds is about –10 W m–2 both at TOA and at

the surface. The magnitude of the low-cloud radiative effect is the largest compared to other cloud

types (except the radiative effects of deep convective clouds at the surface) and is associated with

the largest uncertainty, highlighting the importance of continuous capability for better cloud

observations.

Similarly to clouds, aerosols have a negative direct radiative effect in the SW and a positive effect

in the LW in both all-sky and clear-sky conditions. The total direct effect of aerosol is in the order

of –10 W m–2. The aerosol direct radiative effect does not show a strong variation in latitude;

however, the lack of strong zonal variation is because the effects from different aerosol species

compensate each other, due to changes in aerosol concentration with latitude and due to the

difference in their absorption properties. While our estimates from satellite retrievals generally

agree with those estimated from a global scale, there is a non-negligible discrepancy between our

estimates and those from field campaign data. This discrepancy suggests that better observations

of aerosol absorption properties may be the key element for reducing the uncertainty in aerosol

direct radiative effects.

DACCIWA 603502 20/22


Focusing on outgoing radiative fluxes, models continue to show large biases in both SW and LW,

and larger interannual variability compared to the state-of-the-art satellite observations. While

these persistent errors are highly linked to the ability to model clouds, our analysis shows that

calculations in clear-sky conditions also require caution and improvement, but the sources of errors

tend to be more easily identified compared to all-sky conditions.

The effect of SW irradiance on regional precipitation and circulation are investigated through model

simulations. The SW irradiance is perturbed through increased aerosol concentration and low-

level cloud water amount, leading to different model response. Studies from the DACCIWA

partners demonstrate that model responses can include delayed stratocumulus-to-cumulus

transition, reduced marine flow, more stable boundary layer and reduced convection activity.

Since these responses can be diverse and sensitive to the details of model schemes and setup,

more efforts on the modeling and observational constraints will be needed to confirm how robust

these responses are.

6 References

Allan, R. P. and Ringer, M., 2003. Inconsistencies between satellite estimates of longwave cloud

forcing and dynamical fields from reanalyses. Geophys. Res. Lett., 30(1491).

Bodas-Salcedo, A. et al., 2016. Large Contribution of Supercooled Liquid Clouds to the Solar

Radiation Budget of the Southern Ocean. Journal of Climate, 29(11), pp. 4213-4228.

Cusack, S., Slingo, A., Edwards, J. M. & Wild, M., 1998. The radiative impact of a simple aerosol

climatology on the Hadley Centre atmospheric GCM. Q.J.R. Meteorol. Soc., 124(551), pp. 2517-

2526.

Dee, D. P. et al., 2011. The ERA-Interim reanalysis: configuration and performance of the data

assimilation system. Q.J.R. Meteorol. Soc., 137(656), pp. 553-597.

Deetz, K. et al., 2018. Numerical simulations of aerosol radiative effects and their impact on clouds

and atmospheric dynamics over southern West Africa. Atmos. Chem. Phys., Volume 18, pp. 9767-

9788.

Dewitte, S. et al., 2008. The Geostationary Earth Radiation Budget Edition 1 data processing

algorithms. Advances in Space Research, 41(11), pp. 1906-1913.

Doelling, D. R. et al., 2013. Geostationary Enhanced Temporal Interpolation for CERES Flux

Products. J. Atmos. Oceanic Technol., 30(6), pp. 1072-1090.

Edwards, J. M. & Slingo, A., 1996. Studies with a flexible new radiation code. I: Choosing a

configuration for a large-scale model. Q.J.R. Meteorol. Soc., 122(531), p. 689–719.

Ham, S.-H.et al., 2017. Cloud occurrences and cloud radiative effects (CREs) from CERES-

CALIPSO-CloudSat-MODIS (CCCM) and CloudSat radar-lidar (RL) products. J. Geophys. Res.

Atmos., 122(16), pp. 8852-8884.

Hannak, L. et al., 2017. Why Do Global Climate Models Struggle to Represent Low-Level Clouds in

the West African Summer Monsoon?. Journal of Climate, 30(5), pp. 1665-1687.

Harries, J. E. et al., 2005. The Geostationary Earth Radiation Budget Project. Bull. Amer. Meteor.

Soc., 86(7), pp. 945-960.

DACCIWA 603502 21/22


Hess, M., Koepke, P. & Schult, I., 1998. Optical Properties of Aerosols and Clouds: The Software

Package OPAC. Bull. Amer. Meteor. Soc., Volume 79, pp. 831-844.

Hill, P. G. et al., 2018. Quantifying the Contribution of Different Cloud Types to the Radiation

Budget in Southern West Africa. J. Climate, Volume 31, pp. 5273-5291.

Hill, P. G., Allan, R. P., Chiu, J. C. & Stein, T. H. M., 2016. A multi-satellite climatology of clouds,

radiation and precipitation in southern West Africa and comparison to climate models. J. Geophys.

Res. Atmos., 121(18), pp. 10,857-10,879.

Kato, S. et al., 2011. Improvements of top-of-atmosphere and surface irradiance computations with

CALIPSO-, CloudSat-, and MODIS-derived cloud and aerosol properties. J. Geophys. Res.

Atmos., 116(D19).

Kato, S. et al., 2010. Relationships among cloud occurence frequency, overlap and effective

thickness derived from CALIPSO and CloudSat merged vertical profiles. J. Geophys. Res. Atmos.,

115(D00H28).

Kniffka, A., Knippertz, P. & Fink, A. H., 2018. The role of low-level clouds in the West African

monsoon system. Atmos. Chem. Phys. Discuss., in review.

Knippertz, P. et al., 2015a. The DACCIWA Project: Dynamics–Aerosol–Chemistry–Cloud

Interactions in West Africa.. Bull. Amer. Meteor. Soc., Volume 96, pp. 1451-1460.

Knippertz, P. et al., 2015b. The possible role of local air pollution in climate change in West Africa.

Nat. Climate Change, Volume 5, pp. 815-822.

Knippertz, P. et al., 2017. A meteorological and chemical overview of the DACCIWA field

campaign in West Africa in June–July 2016. Atmos. Chem. Phys., Volume 17, pp. 10893-10918.

Loeb, N., 2008. CERES Level 2 NEWS CCCM Aqua-FM3-MODIS-CAL-CS HDF File - Release B1

[Data set]. s.l.:NASA Langley Atmospheric Science Data Center DAAC.

Mollard, J. D. P. et al., 2018. An assessment of satellite aerosol products in Western Africa during

the Monsoon using DACCIWA observations. In prep..

Muller, C. J. & O'Gorman, P. A., 2011. An energetic perspective on the regional response of

precipitation to climate change. Nature Climate Change, 1(5), pp. 266-271.

Rasch, P. J., Mahowald, N. M. & Eaton, B., 1997. Representations of transport, convection, and

the hydrologic cycle in chemical transport models: Implications for the modeling of short‐lived and

soluble species. J. Geophys. Res. Atmos., 102(D23), pp. 28127-28138.

Stein, T. H. M. et al., 2011. The vertical cloud structure of the West African monsoon: A 4 year

climatology using CloudSat and CALIPSO. J. Geophys. Res. Atmos., 116(D22205).

Taylor, K. E., Stouffer, R. J. & Meehl, G. A., 2012. An Overview of CMIP5 and the Experiment

Design. Bull. Amer. Meteor. Soc., 93(4), pp. 485-498.

Tselioudis, G., Rossow, W., Zhang, Y. & Konsta, D., 2013. Global Weather States and Their

Properties from Passive and Active Satellite Cloud Retrievals. Journal of Climate, 26(19), pp.

7734-7746.

DACCIWA 603502 22/22


van der Linden, R., Fink, A. H. & Redl, R., 2015. Satellite-based climatology of low-level

continental clouds in southern West Africa during the summer monsoon season. J. Geophys. Res.

Atmos., 120(3), pp. 1186-1201.

Waliser, D. E. et al., 2012. The "Year" of Tropical Convection (May 2008 to April 2010): Climate

variability and weather highlights. Bull. Amer. Meteor. Soc., 93(8), pp. 1189-1218.

Wang, J., Rossow, W. B. & Zhang, Y., 2000. Cloud Vertical Structure and Its Variations from a 20-

Yr Global Rawinsonde Dataset. Journal of Climate, 13(17), pp. 3041-3056.

Wielicki, B. A. et al., 1996. Clouds and the Earth's Radiant Energy System (CERES): An Earth

Observing System Experiment. Bull. Amer. Meteor. Soc., 77(5), pp. 853-868.

Yu, H. et al., 2006. A review of measurement-based assessments of the aerosol directradiative

effect and forcing. Atmos. Chem. Phys., Volume 6, pp. 613-666.

Date post:	19-Mar-2021
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Deliverable D5.5 Model evaluation - KIT · Project No: 603502 DACCIWA...

Documents